Previously, the models developed in our laboratory were all low resolution models, with a single pseudoatom representing each nucleotide of the 16S RNA, and a single spherical pseudoatom representing each protein. During the past year, we have begun developing mixed resolution representations for ribonucleoprotein particles. The first advance was to introduce an intermediate resolution protein model, with one pseudoatom per amino acid; this is an appropriate level of detail for representing protein crystal structures in the low resolution models of the ribosome, because it allows accurate treatment of protein shape and orientation. A second advance was the introduction of pseudoatomic representations for transfer RNA (tRNA) and messenger RNA (mRNA). The former is treated as a rigid object whose geometry duplicates that of the tRNA crystal structure, while the latter has a rigid geometry in the region of mRNA/tRNA codon/anticodon interaction and a flexible geometry outside this six nucleotide region. We have also developed automated conversion tools to extract appropriate pseudoatom positions from all-atom representations of RNAs and proteins. These tools enable us to convert atomic models from crystallography, NMR, or detailed all-atom model building to low and intermediate resolution models suitable for YAMMP refinement. They are an essential component of our efforts at developing methods for interconversion between NAB and YAMMP models. Using the advances in models and tools just described, and exploiting the quality and quantity of new data of the ribosome, we are collaborating with J. Frank to develop the next generation models of the ribosome. We have evaluated all the available data, established a suitable set of constraints from the data, and generated preliminary models using the combined distance geometry / YAMMP refinement approach. We expect to complete these second generation efforts during the next year of the grant.